On Sun, Oct 30, 2011 at 6:38 PM, Luke Howard <[email protected]> wrote:
>> mechanism-negotiation mechanism, sort of.  Of course, the way these
>> functions work we'll need something better in the case of SPNEGO being
>> used to negotiate GSS-EAP.  We really need a notion of credentials
>> acquired with credentials, so that we can first acquire a credential
>> handle for GSS-EAP, set what methods it will negotiate, then pass
>> *that* credential handle to another function for acquiring a larger
>> credential set.  In other words, we need a new credential set model
>> for GSS, and the best thing to do for now is to punt.
>
> Could you use gss_add_cred() for this?

Sadly, no.  You'd think, by its name, but it would need *two* input
credential handles to do what we need it to.

I.e., we need a GSS_Add_cred_to_cred_set() that takes one credential
handle that is to be an element of the other.  We'd then need not
GSS_Set_neg_mechs() but GSS_Set_net_creds().

I.e., we want credential handles as sets of credential handles, with
credentials for mechanism-negotiation mechanisms themselves acting as
sets (nested sets).

> Myself, I find gss_acquire_cred() + gss_set_cred_option(EAP_METHODS) easier 
> to understand, particularly if this kind of application-level configurability 
> is an edge case (I really want applications not to have to worry about this; 
> rather, they use SPNEGO, and the mechanism and/or its credential manager 
> sorts things out for you).

GSS_Add_cred() was a mistake :(

We may need a wrapper around all this to make GSS dead simple for most apps.

Nico
--
_______________________________________________
abfab mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/abfab

Reply via email to